Against Quietist Normative Realism Forthcoming in Philosophical Studies

Tristram McPherson University of Minnesota Duluth [email protected]

Note: This is the author’s penultimate manuscript of this paper (ms. of March 1, 2010). The final publication is available at www.springerlink.com.

Abstract:

Recently, some philosophers have suggested that a form of robust realism about ethics, or normativity more generally, does not face a significant explanatory burden in metaphysics. I call this view metaphysically quietist normative realism. This paper argues that while this view can appear to constitute an attractive alternative to more traditional forms of normative realism, it cannot deliver on this promise. I examine T. M. Scanlon’s attempt to defend such a quietist realism, and argue that rather than silencing metaphysical questions about normative reasons, his defense at best succeeds only in shifting the focus of metaphysical enquiry. I then set aside the details of Scanlon’s view, and argue on general grounds that that the quietist realist cannot finesse a crucial metanormative task: to explain the contrast between the correct normative system and alternative putatively normative standards.

Keywords: metaethics, metanormative quietism, metanormative realism, normativity, reasons, Scanlon

McPherson Against quietist normative realism MS 1

Introduction

Philosophers interested in normative domains such as ethics or epistemology face a persistent challenge: to understand how our practices of normative judgment and discourse fit within our best general account of the world. I will call this the metanormative project.1 (Contrast the more familiar metaethical project, which addresses similar questions about specifically moral or practical norms.) Roughly, metanormative realists think that normative claims are made true by their correspondence to the normative facts.2 Metanormative realists appear to face a daunting metaphysical challenge, which can be partially characterized by noting three central desiderata for a metanormative theory. The first desideratum is to fit with our broader metaphysical commitments. The second is to permit us to explain how normative facts are epistemically accessible, as the non-skeptical realist presupposes them to be. The third is to permit us to explain the distinctive authority that we presume normative claims to have in our deliberation. Some philosophers have recently suggested that the appearance of such a challenge is illusory. They claim that while the realist is correct to take there to be normative facts, questions that might appear to require metaphysical answers can instead be addressed by substantive normative theorizing.3 They thus suggest a sort of metaphysically quietist normative realism. For brevity, I will call this view quietist realism. I begin this paper by distinguishing quietist realism from a series of other metanormative theories, and explaining why it appears to be a promising alternative (§1). I argue in the rest of the paper that quietist realism cannot deliver on this promise. I examine T. M. Scanlon’s attempt to defend quietist realism, and argue that rather than silencing metaphysical questions about normative reasons, his defense at best succeeds only in shifting the focus of our metaphysical enquiry (§2). I then set aside the details of Scanlon’s view, and argue that the quietist realist cannot finesse a crucial element of the third explanatory desideratum identified above. This is the demand to explain what underwrites the contrast between the correct normative system and other formally normative standards (§3).

1 I take the term ‘metanormative’ from Enoch 2007. Compare Hussain, 2004, 150-1 for a slightly different interpretation of the central project. 2 I oversimplify: one might (à la Quine) reject an ontology of facts on general grounds, but nonetheless remain a realist about some domains. I set aside this complication. 3 Besides Scanlon, who will be my central exemplar of this view in §2, philosophers who have made remarks that may be suggestive of such a view include Parfit “Appendix A: Normativity Naturalism, and Non-Cognitivism.” (Ms. of June 2008), Nagel 1997 Ch.6, and perhaps Dworkin 1996. Parfit’s views on these matters are still evolving, and the interpretation of Dworkin and Nagel in this respect is controversial. On Dworkin, compare Zangwill 1997 and Dworkin’s 1997 response. Svavarsdottir 2001 is an extremely helpful exploration of Nagel’s views on these topics. For an important related view, see Kramer 2009. McPherson Against quietist normative realism MS 2

1. Quietist realism distinguished and motivated

Quietist realism is characterized by two claims. On the one hand, it is a form of realism, accepting that there are normative facts and properties. On the other, it suggests that accepting the existence of such facts and properties does not lead to the sort of explanatory burdens mentioned in the Introduction. Quietist realism of this sort is not typically discussed in the metaethical literature, so I will begin by distinguishing it from more familiar positions. First, standard versions of naturalist normative realism claim that there are normative facts and properties, and offer accounts of those facts as part of the natural order.4 Typically they do this either by proposing a reduction of normative properties to other natural properties, or by arguing that normative properties, while not reducible, are metaphysically continuous with other natural properties. Such naturalist views typically seek to explain how those facts satisfy the sorts of desiderata identified in the Introduction. One common complaint about such naturalist forms of realism is that, despite claims to the contrary, they fail to adequately explain the normative authority of ethical or epistemic facts, thus failing to satisfy the third desideratum.5 Recently, some normative realists have taken a deflationary tack, arguing against this desideratum by denying that normative authority is a coherent idea, if construed as being anything other than the characteristic of any standard against which one can make a mistake.6 Second, on standard expressivist views, normative attitudes are at the explanatorily fundamental level ‘desire-like’: their functional role is to have the world conform to their contents, rather than for their content to conform to the world.7 Leading expressivists about normative thought and discourse have recently allied their view with a minimalist account of truth and fact, according to which they can correctly claim that there are normative truths and facts.8 Such quasi-realist views contrast with normative realism, because according to the quasi-realist the philosophically fundamental explanation of the relevant talk of ‘facts’ and ‘truths’ is attitude-expressive rather than world-reflective.9

4 Leading examples of metanormative or more narrowly metaethical naturalist realists include Boyd 1997, Railton 1997, Smith 1994, Jackson and Pettit 1995, Schroeder 2007. 5 Complaints of this kind directed at specific naturalist accounts of ethics and epistemology are ubiquitous. Parfit’s “Appendix A: Normativity Naturalism, and Non-Cognitivism.” (Ms. of June 2008) offers an unusually strong general version of this charge against the naturalist. 6 This view is raised as a serious possibility by Copp 2004 and defended by Tiffany forthcoming. 7 Helpful discussions of the idea of direction of fit sketched here include Humberstone 1992 and Velleman 1992. 8 See for example Gibbard 2003, and especially Blackburn 1993. 9 It is not trivial to spell out this idea once one accepts minimalism. See Dreier 2005 for a helpful discussion. Quasi-realism should be distinguished from the sort of view that a normative realist might have if she found the expressivist treatment of the distinctive normativity of practical discourse compelling, and sought to graft that account onto a realist framework (for attempts at such grafting for different purposes, see Copp 2001 and Finlay 2005). Quietist realists show no signs of seeking to deploy such a strategy, so I set it aside here. McPherson Against quietist normative realism MS 3

Third, traditional non-naturalist realism about normativity insists that normativity is a sui generis property, distinct in kind from any of the fundamental properties that explain the non-normative nature of reality.10 Non-naturalists often claim that their view is the only one that can explain the robust normativity of ethics or epistemology. However, this view also faces serious challenges, most notably with respect to the first and second desiderata mentioned above. For example, it has been argued that dusting our ontology with brute normative properties is ontologically profligate, and that the non-naturalist cannot explain the apparent ‘ban on mixed worlds:’ the fact that we think that there are no ways that the world might be that are non-normative identical but normatively distinct.11 Finally, rather than seeking to address the desiderata on metanormative theories, as the above accounts each do, one might insist that the demand to do so is unreasonable given our epistemic capacities. Thus, one could claim that, while there are meaningful and substantive metaphysical questions about the nature of normativity, we are not capable of offering positive answers to those questions. This would be a kind of mysterianism about the nature of normativity.12 The mysterian idea runs counter to the explanatory aspirations that typically animate philosophical enquiry. Indeed, metanormative mysterianism has been less often explicitly embraced than it has been lodged as a complaint against non-naturalist realists by their opponents.13 Quietist realism shares with each of these views a commitment to normative truths and facts. However, it is otherwise distinct from each of them. Against the deflationary realist, the quietist rejects the idea that there is nothing to normativity but the formal property of being a standard that one can fail to meet. Against the quasi-realist, he rejects the idea that, at the fundamental explanatory level, normativity is to be explained in terms of attitudes. However, the quietist realist shares with the quasi-realist the thought that the traditional naturalists and non- naturalists, and the mysterian, misunderstand the sort of desiderata mentioned as requiring a metaphysical solution. The quietist’s general strategy is to suggest that the relevant explanatory burdens are best understood as falling on, and being met by, substantive normative theories rather than metanormative theories. This strategy raises important questions concerning the distinction between normative and metanormative theories. As they are typically understood, normative theories are systematic accounts of the structure and content of the correct normative evaluation of agents, states of affairs, and actions. Thus, normative theorists offer accounts of the sorts of considerations that count as reasons for action or belief, the sorts of actions that are right and wrong, the conditions that amount to

10 Compare Moore 1903, Shafer-Landau 2003, and Fitzpatrick, 2007, among many others. 11 The classic statement of this charge comes in the papers reprinted as chapters 6 and 7 in Blackburn’s 1993. For an important recent response, see §s 8-11 of Fitzpatrick 2007. 12 The term ‘mysterianism’ is most commonly used in the philosophy of mind, to label the view that the relationship between experiential states and underlying physical causes necessarily eludes our epistemic grasp (compare McGinn 1993 and Stoljar 2006, 91-2). 13 For example, this is one thread of Korsgaard’s critique of ‘substantive moral realists’: she concludes that such realism “…refuses to answer the normative question. It is a way of saying that it cannot be done” (1996, 39). McPherson Against quietist normative realism MS 4

knowledge or epistemic justification, etc., and of the relationships between these different normative dimensions. If normative theories do all of this, what is left for metanormative theories to do? Briefly, such theories provide an interpretation of the content of the fundamental claims of a normative theory. For example, consider a very simple act utilitarian normative ethic, which claims that an action is wrong just in case and because it will bring about less net happiness than some other option available to an agent. This theory can be interpreted in a variety of ways. A reductive realist might identify the wrongness of an action with the sub-optimality of its outcomes. A non- naturalist might insist that wrongness is a distinct sui generis normative property that nonetheless necessarily follows from the sub-optimality of its outcomes. An expressivist might say that to sincerely avow the simple utilitarian ethic is just to express a distinctive attitude (moral disapproval) towards any action with suboptimal outcomes. These metanormative theories are competing interpretations of how to fit this simple normative theory into our broader account of the world.14 Quietist realism is distinctive in arguing that the correct interpretation has much less to explain than is presupposed by these other theories. Quietist realism is a seemingly attractive metanormative view for at least two reasons. First, a central motive for metanormative anti-realisms is the sense that the apparent desiderata on realist metanormative theories simply cannot be jointly satisfied. If these apparent desiderata are in fact misunderstood when treated as constraining the metaphysics of normativity, as quietist realism suggests, realism will turn out to be a much easier metanormative view to defend. Second, by insisting that explanatory questions about normativity be answered at the normative level rather than the metanormative level, the quietist realist expresses commitment to a potentially attractive explanatory autonomy for the normative. This sort of explanatory autonomy in turn promises to vindicate the priority of substantive normative theorizing in ethics and epistemology. This priority can appear to be quite robust, licensed both by general dicta like Hume’s Law (the claim that one ‘cannot derive an ought from an is’) and by the actual practice of reasoning and theorizing in ethics and epistemology.15 Quietist realism thus appears to be an attractive candidate metanormative theory. However, I will argue that it cannot deliver on this promise. I begin by considering Scanlon’s attempt to defend quietist realism.

2. Scanlon’s arguments for quietism

In What We Owe to Each Other, Scanlon presents a powerful contractualist account of the morality of right and wrong. He argues that the wrongness of an action is to be explained as a function of a distinctive class of normative reasons (1998, esp. Ch. 4).

14 For a more detailed discussion of the distinction between metaethics and normative ethics see McPherson 2008a, ch. 1.For worries about the distinction between metaethics and normative ethics see Kagan 1998, Dworkin 1996, and Korsgaard 2003. 15 For further discussion of a strong version of such justificatory autonomy in ethics, see McPherson 2008b. McPherson Against quietist normative realism MS 5

On this account, normative reasons are the fundamental normative entities. (Contrasting views would make virtuous character traits, or constitutive rules for action, or the value of certain outcomes normatively fundamental.) Scanlon claims that by identifying these distinctive reasons, his account vindicates the importance and priority that we typically grant to judgments of right and wrong in our reasoning. Scanlon also claims that if this project is successful, “[n]o interesting question would remain about the ontology of morals – for example about the metaphysical status of moral facts” (1998, 2). Because his strategy is to explain morality in terms of normative reasons, this suggests that, when properly understood, commitment to the existence of normative reasons does not raise any important metaphysical questions. As we saw in §1 above, the quasi-realist can escape metaphysical questions about normativity by appealing to an expressivist account of normative claims. However, Scanlon rejects this strategy, claiming against the expressivist that judgments about reasons express beliefs (1998, 59 and 2003, 7). He also claims that normative beliefs are not ‘about the natural world’ (1998, 60).16 Scanlon is thus a sort of non-naturalist realist about reasons. As we saw in the Introduction, metanormative realism appears to face a series of explanatory challenges. Scanlon suggests two main responses aimed at undercutting this appearance.17 He first claims that reasons are ontologically unproblematic because they are simply a class of propositions. His second and more important response generalizes his strategy for understanding morality. The latter strategy explains wrongness in terms of another normative notion: reasons. Scanlon suggests that apparently metaphysical doubts about reasons are likewise amenable to a normative explanation. The upshot of his discussion is a form of what I have called quietist realism: while he accepts a sort of non-naturalist realism about normative reasons, he takes himself not to face the sort of explanatory challenges in metaphysics that I identified in the Introduction.

16 Scanlon glosses natural facts as including “…all and only the physical and (insofar as they may be different) psychological facts” (2003, 8). He also suggests the different idea that the natural world is just the world as described by science. The distinction between natural and non-natural properties is actually quite vexing. I set aside the very real interpretive issues here, which are orthogonal to the argument of this section. 17 As we noted in §1, non-naturalism is often taken to make the demand for metaphysical explanation especially intense. This is in part because the non-naturalist claims that the sorts of properties that explain normativity are different in kind from the sorts of properties identified by the sciences. Scanlon argues that on his conception, reasons do not conflict with the picture of the world offered by science because plausible claims about reasons do not entail predictions about the spatiotemporal world that might conflict with those generated by scientific theories (2003, 7-9). This response assumes (implausibly, it seems to me,) that the ‘picture of the world’ offered by contemporary science is limited to specific positive ontological claims made by scientific theories, and does not involve more general commitments concerning what kinds of things exist. I set this concern aside here. McPherson Against quietist normative realism MS 6

Scanlon’s first response appeals to the view that reasons are just propositions in order to show that there are no distinctive ontological questions about reasons.18 For example, on this account one reason would be: POKE Poking Artemis with a stick would cause her to suffer This response, however, fails to address the central ontological question about reasons, for reasons that Scanlon himself should accept. Recall that Scanlon is a non- naturalist about reasons. However, this commitment surely does not rest on a commitment to the impossibility of naturalistic reduction of propositions. The metaphysical issues about reasons instead concern a class of relational facts: facts that identify certain propositions as reasons for certain actors to have certain responses.19 Consider an example of the relevant sort of relation: NORM POKE is a reason for anyone to refrain from poking Artemis The interesting metaphysical issue here does not concern the ontological status of NORM itself, which is again simply a proposition. Rather, it concerns what makes NORM true. After all, the following also expresses a proposition: NORM* POKE is a reason for anyone to poke Artemis I take it that NORM is true and NORM* is false. The normative realist needs to explain this contrast.20 Scanlon’s realism thus requires the satisfaction of is-a-reason- for relations like NORM, and the non-satisfaction of relations like NORM*. Our metaphysics must therefore seemingly contain relational structures that make NORM true, and lack those that would make NORM* true. The propositional account of reasons (if it is correct) thus serves only to clarify the locus of metaphysical questions about normativity, not to provide an easy answer to them. Scanlon’s second response to metaphysical worries about reasons is more complex. He first identifies a series of facts about reasons that might seem to require metaphysical explanation. These include the fact that claims about reasons can be true or false, the fact that we can be mistaken about them, and the fact that they seemingly must be about some subject matter independent of us (1998, 62). As a non-naturalist, Scanlon does not seek to explain these facts in purely naturalistic terms. However, he also denies that ‘Platonist’ metaphysical commitments are required to underwrite them. Rather, he suggests that “…what is necessary is for there to be standards for arriving at conclusions about reasons” (1998, 63). In an evocative but opaque image, he suggests that to call moral claims true is thus not to

18 1998, 57. One might worry over whether we should let propositions into our ontology, and hence whether, on the propositional account of reasons-talk, we should admit reasons. For reasons that will become clear immediately below, I set this concern aside. 19 Scanlon himself notes that the ‘element of normative force’ does not adhere to propositions themselves, but rather to judgments that if a proposition were true, it would be a good reason for a certain action or belief (1998, 57). 20 A similar point is briefly made by Finlay 2007, 840. Compare also Olson’s related 2009, §3 discussion of what he calls the ‘extensional fallacy.’ McPherson Against quietist normative realism MS 7

grant them “‘intrinsic metaphysical gravitas’” (2003, 9; emphasis and scare quotes his).21 Rather, he suggests, it is to grant them what he calls ‘normative gravitas’. An example inspired by Scanlon’s own use of mathematics as an analogy may help to clarify the idea deployed here. Famously, questions like “is there an even prime?” seem to call for answers that commit us to the existence of numbers. And this has often been taken to raise vexing ontological questions in the philosophy of mathematics. Scanlon might reply that whether there is an even prime is a mathematical question, to be settled simply by applying correct standards of mathematical reasoning. It is thus not a metaphysical question to be settled by appeal to metaphysical standards external to mathematics itself. Similarly, the fact that one reason not to poke Artemis with a stick is that doing so would cause her to suffer appears to commit us to an ontology of normative facts – facts capable of explaining why NORM is true and NORM* is false. However, Scanlon suggests that we can explain such contrasts by appealing to standards of correctness for practical reasoning, which are themselves normative standards (1998, 63).22 This discussion appears to show how some seemingly metaphysical questions can be shown to be normative. However, the structure of Scanlon’s reply suggests a natural question: might the same metaphysical concerns arise about standards of correctness? Scanlon responds that claims about these standards are also normative and not metaphysical. More specifically, he says the following: The question of whether there are standards of the required sort is a substantive one within the subject in question – a matter of whether there are conclusions and ways of arriving at them that we have no reason to regard as defective. (1998, 63) This explanation is inadequate as written for two reasons. First, on a natural reading, it is formally objectionable. On this reading, the existence of standards of reasoning follows from the existence of conclusions (and ways of arriving at them) that we have no reason to regard as defective. But suppose for reductio that error theory about reasons-talk were true (compare Mackie 1977). This would entail that there were no reasons, and hence no reasons to regard anything as defective. Scanlon’s explanation, however, takes this to be sufficient for the existence of reasons. On a natural reading, Scanlon’s explanation thus suggests absurdly that such an error theory about reasons would secure the existence of standards of reasoning.

21 Scanlon takes this language from Wright 1996, 5. One might worry that this appeal undermines Scanlon’s realist pretensions. Roughly, Wright suggests that we might earn the right to talk of truth and falsehood within a domain of discourse in virtue of the internal discipline of the discourse, rather than by appealing to a more ‘metaphysical’ basis for the truth of such claims. Wright conceives of this denial of ‘metaphysical gravitas’ as anti-realist in an important sense, and with good reason: there is nothing in the idea of an internally disciplined discourse that suggests that there couldn’t be a variety of inconsistent but equally disciplined discourses, which our deliberative practices might just as easily have instantiated. 22 I take it that Scanlon’s commitment to the idea that standards for correct reasoning are fundamental helps to explain his otherwise surprising claim that his view is, in Korsgaard’s (1996, 34-37) terms, a form of ‘procedural’ as opposed to ‘substantive’ realism (Scanlon 1998, 380 n.48). However, Scanlon’s discussion of Korsgaard in 2003, 13-15 suggests that he may no longer want to characterize his view in this way. McPherson Against quietist normative realism MS 8

Second, what conclusions or methods we have reason to regard as defective is a function of what evidence we possess concerning these matter. Because of this, Scanlon’s formulation has implausible implications about the significance of misleading evidence. For example, it seems possible that reasons could exist, while our misleading evidence entails that we ought to suspect that all of our methods and conclusions are defective.23 Because Scanlon suggests that the existence of standards is ‘a matter of’ what we have reason to think, his explanation thus suggests implausibly that the nonexistence of normative standards would follow from a preponderance of a certain kind of misleading evidence. The first problem with the quoted explanation is thus that it appeals to the nonexistence of certain reasons, and the second is that it appeals to an evidential condition. Revising the explanation to remove these elements would produce the following principle: Revised there are standards for arriving at conclusions about reasons just in case and because there are conclusions and ways of arriving at them that the balance of objective (i.e. not evidence-relative) reasons supports treating as non-defective. Accepting this revision would lead Scanlon into an objectionably tight explanatory circle. Revised suggests that facts about standards of reasoning are to be explained in terms of facts about reasons. However, as we saw above, Scanlon sought to explain the truth of claims about reasons not in metaphysical terms, but by appeal to standards of reasoning. Different normative theories give fundamental explanatory priority to different notions: reasons, values, deliberative standards, etc. Accepting one of these theories implies commitment to being able to give normative-level explanations of facts involving the non-fundamental notions in terms of the fundamental normative notions. Scanlon’s strategy for reinterpreting apparently metaphysical questions about reasons as themselves normative questions attempts to generalize this explanatory move. However, the asymmetry implicit in the explanatory pretensions of these normative theories vitiates his attempt. Reasons and standards for reasoning cannot very well each have explanatory priority over each other. One might resist the idea that the explanatory circularity identified here is vicious by suggesting an analogy to epistemic or semantic holism. On this picture, claims of explanatory priority between normative notions like reasons and standards of reasoning should not be taken seriously. Instead, the truth of any particular normative claim (whether about reasons or standards) is to be explained by its relation to the system of normative considerations as a whole. This suggestion however, still does not suffice to eliminate or ‘normativize’ apparently metaphysical questions about reasons. This is because we can raise these questions about the normative system as a whole.24

23 Such evidence would be deeply peculiar in virtue of being self-undermining: if one had it, one could not coherently treat it as evidence. However, since one can have evidence that one does not recognize as such, this does not vitiate the objection. 24 For example, Jackson and Pettit 1995 offer a holistic account of moral semantics that they conjoin with a naturalistic metaphysics of morality. McPherson Against quietist normative realism MS 9

The analogy to semantic holism is instructive here. According to semantic holism, what I mean by a particular word is determined by place in the whole structure of my idiolect or the linguistic structure of my community. A semantic holist can thus deny that the meanings of particular words can be determined in isolation from that of their fellows, or that some concepts are semantically fundamental atoms in terms of which the meanings of all words are constructed. This sort of semantic holism would (if correct) provide important constraints on the possible nature of the correct metaphysical account of meaning. However, such holism is entirely compatible with (for example) a naturalistic reduction of facts about this whole system of meanings to some combination of patterns in the use of linguistic tokens, the dispositions of language users, and the causal connections between those language users and their environment. If correct, the sort of holistic view of semantics just sketched does not foreclose metaphysical questions about meaning. Rather, as we have just seen, it helps us to more precisely formulate those questions. If we interpret it by analogy to such semantic holism, Scanlon’s strategy would thus fail to quiet genuine metaphysical questions about normativity. Like his appeal to the propositional account of reasons, his discussion of standards for reasoning may at best prompt us to think about whether we should direct such metaphysical questions at some fundamental normative notion or at a normative system as a whole. In this section, I have illustrated the quietist realist’s general strategy by examining Scanlon’s defense of the view. I argued that this defense fails to make real progress in showing how the relevant apparently metaphysical questions can be answered within a normative theory. However, it would be premature to abandon a theory as intriguing as quietist realism simply because the defense offered for it fails: perhaps a better argument remains to be made. In the next section, I offer a more general argument against quietist realism, focusing on the burden of explaining the distinctive authority that we take normative claims to have in our deliberations.

3. Quietism and normative asymmetry

As I noted in the Introduction, one desideratum on a metanormative theory is that it vindicate the authority that epistemic and practical norms appear to possess in our deliberations. This desideratum can be restated in contrastive terms: we want to be able to explain what makes one putative normative standard authoritative in our deliberations, while another standard is not. In this section, I first clarify this desideratum and explain why the quietist realist is committed to meeting it. I then argue that quietist realism fails because it cannot accommodate this desideratum. To begin, consider a pair of normative standards where this contrast does not arise. Facts about the legality of chess moves have what I will call formal normativity. Chess is formally normative simply in virtue of its being possible to play McPherson Against quietist normative realism MS 10

an incorrect chess move.25 For example, moving one’s Knight diagonally is incorrect relative to the norms of chess. Next consider an extremely similar game which I shall dub schmess. Schmess is identical to chess except that in schmess one is permitted to move one’s Knight diagonally. Like chess, schmess is formally normative: for example, moving one’s Knight along a rank or file violates both the rules of chess and those of schmess. For simplicity, call the correct norms on action the reasons standard.26 The reasons standard is also formally normative, because it is possible to act contrary to your reasons. For example, one would violate the reasons standard by poking Artemis in circumstances in which there are decisive reasons not to do so. Just as we can distinguish the rules of chess from those of schmess, we can distinguish the reasons standard from an alternative standard for action, which I will call the schmeasons standard.27 To focus ideas, suppose that the schmeasons standard is ‘disutilitarian’: one has schmeason to A just to the extent that A-ing will increase the net suffering in the world. The schmeasons standard is also formally normative. The kind-hearted, for example, regularly fail to act in accordance with their schmeasons. Each of these four standards (chess, schmess, reasons and schmeasons) has the three features that Scanlon noticed might seem to call for a metaphysical explanation. For example, claims about the legality of a chess move can be true or false, we can be mistaken about whether a move is legal, and claims about chess concern a subject matter independent of us: chess. By contrast, consider Calvinball (from the comic strip Calvin and Hobbes), whose rules are whatever Calvin decides they are at any given time. Calvinball fails the last two of the above criteria, at least relative to Calvin himself. The fact that all four standards share the features that Scanlon identifies as exemplary explananda suggests that even if he had been successful in offering a purely normative explanation of them, he would not have offered a quietist account of what is distinctive of reasons. Almost all of us take reasons to be normative in a way that contrasts with other formally normative systems such as the rules of chess or the schmeasons standard.28 The third desideratum for metanormative theories that I identified in the Introduction was to explain this distinctive authority that we presume the reasons standard to have. To give the problem a pithy label: metanormative realists need to explain why reasons are robustly normative, while schmeasons are merely formally normative. The quietist realist cannot afford to downplay or reject this desideratum. This is because one of the main reasons why philosophers reject naturalist realism is

25 Compare Copp 2004’s similar (but slightly narrower) notion of generic normativity, and Foot 1972 on the normativity of etiquette. I set aside the other central dimension of chess normativity, which is illustrated by the fact that it is possible to play a legal but bad move. 26 Here and below, I talk for the sake of brevity as if Scanlon is correct that reasons are the fundamental normative notions. Nothing hangs on this: for example my argument could be translated without loss into a discussion of values, virtues or standards for reasoning. 27 The language here is inspired by Enoch’s “Agency, Schmagency…” (2006). However, I am pursuing a different point with the contrast than Enoch is with his. 28 Again, the deflationary naturalism defended by Tiffany forthcoming denies this asymmetry. McPherson Against quietist normative realism MS 11

that they take it to be impossible for naturalist theories to meet this desideratum.29 Giving it up would thus leave quietist realists dialectically impotent against the sort of reductive strategy suggested by Mark Schroeder (2007, 79-82). Schroeder claims that he can do everything that is required to ‘capture’ normativity simply by arguing that normativity is fundamentally about reasons (as Scanlon, for example, would agree), and then offering a reductive account of reasons. If Schroeder can offer an extensionally plausible account of reasons, the non-naturalist’s central remaining objection to his view must concern whether robust normativity could ultimately amount simply to being desired. It is thus unsurprising that we see quietist realists like Scanlon deploying this desideratum against their naturalist opponents. For example, Scanlon rejects the theory that reasons for action are reducible to second-order desires in the following way: But if second-order desires are really desires, then there is the question of how their second-order character, if it is just a difference in the objects of these desires, can give them the kind of authority that is involved when one reason supports the judgment that another putative reason is in fact irrelevant. (1998, 55; emphasis in original) Scanlon’s complaint here is not extensional: it is not that the second-order desire account is intuitively incorrect concerning which reasons we have. Rather, Scanlon finds a desire’s being second-order too arbitrary a feature to form the explanatory ground for the normative authority of a consideration.30 Quietist realism thus appears committed to the possibility of explaining the difference between robust and merely formal normativity. Note that in order to do so, it would have to satisfy at least three conditions. First, the correct normative standard would need to have a feature (call it F) that explained the robust normativity of that standard. Second, F must be non-detachable from each substantive norm within the correct normative standard. For if it were not, there could be a normative standard that was inconsistent with the correct normative standard, but that was nonetheless just as robustly normative, in virtue of having F. This would in turn entail that there would be no non-arbitrary grounds for choice between the correct normative standard and this alternative. Third, F must not simply reflect the metaphysical nature of normativity, for to do so would be to abandon the quietist program.31 Thus, the non-detachability at play must be

29 Parfit MS §15 goes so far as to suggest that naturalist realists simply could not be talking about the same sort of thing that he is interested in when he makes normative claims, because such views cannot coherently explain the robust normativity of normative properties. 30 It is notable that Moore’s 1903, §13 complaint about the second-order desire view as an analysis of goodness appears to focus entirely on the intuitive case against synonymy. Contemporary opponents of naturalism have tended to abandon the conceptual interpretation of the open question argument as insufficient. An especially clear example is Rosati 1995, which suggests that the open question argument can be resuscitated exactly by focusing on the alleged intuitive inadequacy of naturalist attempts to explain normative authority. 31 Consider two examples of such a metaphysical explanation. According to the reductive second- order desire theory of reasons, to be robustly normative just is to be properly related to a second- order desire. By contrast, the traditional non-naturalist insists that robust normativity is a sui generis McPherson Against quietist normative realism MS 12

somehow ‘internal’ to the normative itself. In order for the quietist strategy to satisfy these three conditions, there must be something that unifies all reasons and imparts normativity to them, and that is not a possible feature of any alternative putative normative system, all while not being a metaphysical property or complex of properties. The most prominent view that can be seen as seeking to satisfy these desiderata is the ambitious Kantianism developed most famously by Christine Korsgaard. Korsgaard argues (a) that normativity should be understood in terms of the transcendental conditions on agency, (b) that substantive normative principles can all be derived from these conditions, and (c) that the normativity of these conditions is only directly available from the perspective of someone who must act, and not from the theoretical perspective (see both her 1996, and especially her 2003). If they are defensible, these three elements could help to meet the constraints just introduced. Thus, thesis (c) might enable us to explain why the account of the normativity of reasons must arise internal to the normative perspective and not as an extrinsic metaphysical supposition. Theses (a) and (b) together purport to explain why there is only one robustly normative standard. The interpretation and defensibility of each of these elements of Korsgaard’s view are intensely controversial. For example, Scanlon appears to reject all three (2003, 14-17). Further, Korsgaard has suggested that her view somehow transcends the distinction between realists and anti-realists entirely (this is a central theme of her 2003). Thus, while Korsgaard’s view provides the leading model of a non- metaphysical attempt to satisfy these constraints, neither its failure nor the correctness of her realism-transcending interpretation of its structure would suggest hopeful prospects for quietist realism. Scanlon’s argument discussed in §2 above suggests the general form that the quietist’s strategy to meet this desideratum must take: to offer a normative-level explanation of robust normativity. This use of the strategy still faces the diagnosis that I offered for the failure of Scanlon’s use of this strategy in §2. My diagnosis suggested that whatever normative-level explanation one offers, further metanormative interpretations of that explanation are still possible. This is an intrinsic problem for quietist realism, according to which such interpretive questions must be ill-posed. However, this diagnosis leads to two further and decisive problems for the quietist explanation of the robust normativity of reasons. Consider first an apparently simple way of drawing the relevant contrast between reasons and schmeasons from within the normative perspective. It begins by noting that one makes a mistake relative to the reasons standard in treating POKE as a reason to poke Artemis. Generalizing, it is a mistake relative to the reasons standard to treat schmeasons as reasons. One might argue that this is enough to explain the asymmetry in robust normativity between reasons and schmeasons. There is undeniably such a contrast in how reasons and schmeasons are evaluated within the reasons standard. However, this contrast cannot explain the metaphysical property. On either of these views, the explanation of the contrast between reasons and schmeasons is made by appeal to the metaphysical nature of normative properties, and is thus inconsistent with quietist realism. McPherson Against quietist normative realism MS 13

asymmetry in robust normativity between reasons and schmeasons. This is because the same pattern applies to chess and schmess: as we have seen, one plays chess incorrectly if one moves one’s Knight diagonally, despite doing so being legal in schmess. However, chess is paradigmatically a non-robust normative standard. Because this sort of asymmetry occurs within any formally normative standard, it cannot explain the robust normativity of reasons. Thus, the attempt to appeal to correctness relative to the reasons standard at best presupposes that this standard is robustly normative, rather than shedding any light on its normativity. One might still hope that the quietist strategy could be made to work in virtue of the fact that epistemic or practical norms are much richer in content than the rules of chess.32 However, the same problem will afflict other attempts to explain robust normativity by appeal to structural features within the reasons standard. This is because any structural feature of the reasons standard could seemingly be detached and thus could be shared by structurally analogous but non- robust norms. Call this the problem of structural parallelism. This suggests that we need something external to a normative perspective itself in order to explain whether and why that perspective is robustly normative. The second problem facing the quietist realist is that some metanormative theories describe norms that are obviously not robustly normative, despite having normative implications that are identical to the content of the correct normative theory. Call this the problem of normative coincidence. This is a problem for quietist realism, because if robust normativity could be explained internally to the correct normative theory, as quietist realism requires, normative coincidence would be impossible. The problem of normative coincidence arises from a familiar point from the history of ethical theory: that it is possible for two people to agree on a normative theory while disagreeing on the correct metanormative account.33 For example, various utilitarians have espoused expressivist, naturalist realist, and non-naturalist realist metaethical views. The problem can be brought out by an inverted ‘Euthyphro’ example that exploits this general point. Consider again the reasons standard, understood (by hypothesis) as the correct normative ethical theory. Suppose that Hades is capable of understanding all of the propositional content of this theory. Suppose finally that Hades is briefly in a really bad mood, and comes momentarily to hate all and only what is approved of by the theory. Now consider the metanormative theory on which to be a reason is just to be a consideration that Hades hates34 at that very moment.35

32 I am grateful to Sari Kisilevsky for pressing this line of objection. 33 This claim has been explicitly made across the history of discussions of metaethics and normative ethics. For example, compare Frankena 1951, 45, Zimmerman 1980, 659, and Kagan 1998, 5 (Kagan does not explicitly endorse this claim). Sumner 1967, 96ff shows how a stronger version of this sort of claim has sometimes motivated the thought that metaethics must be ‘normatively neutral’. 34 I oversimplify: the metanormative theory would also need to preserve the structure of the correct normative theory. So if the correct normative theory were a virtue theory, then on this theory to be a virtue would be to be a character trait hated by Hades, and Hades would need to hate actions because they expressed virtue, etc. The task of filling out the simplified story given in the text is structurally analogous to the task of developing an intuitively extensionally adequate Divine Command metaethic. McPherson Against quietist normative realism MS 14

A cousin of Scanlon’s objection to the second-order desire theory looks powerful here: whatever Hades hates, Hades’ happening to hate something at a particular moment is too arbitrary a feature to constitute a robustly normative standard. However, the Hades-hatedness theory is by hypothesis normatively identical to the reasons-standard, and any quietist realist account will thus entail that it is just as robustly normative as the reasons-standard.36 The quietist realist thus entails implausibly that any metanormative account with the right normative shape – even Hades-hatedness – is robustly normative. The quietist might seek to reply to the problem of normative coincidence by arguing that despite their being necessarily coextensive, the property picked out by the reasons standard is somehow normatively prior to the property of being (rigidly) Hades-hated.37 The quietist needs the claim of normative priority to itself be a normative claim. But if it is, then it is compatible with the truth of Hades-hatedness, which, by hypothesis, is normatively coextensional with the correct normative theory. To see this more clearly, consider a more familiar sort of example: suppose for the moment that the correct normative theory is utilitarian, so that facts about the overall level of happiness are normatively fundamental. The divine command theorist says: this is true because God commands us to be sensitive to this property. The divine command theorist can and must deny that this is a normative claim. Rather, this is a claim that purports to explain why certain properties are normative.38 The problem of structural parallelism and the problem of normative coincidence each suggest that attempted normative-level explanation of robust normativity would inevitably over-generate robust normativity in unacceptable ways. My diagnosis in §2 suggested that different metanormative interpretations of a given normative theory are always possible. The two problems just identified suggest that such interpretation is also essential to providing a defensible account of robust normativity. I have argued that the quietist realist cannot hope to answer the question “in virtue of what is one standard robustly normative while another standard not?” by appealing solely to features internal to a normative perspective. This suggests that the only remaining option for the quietist realist is to deny that this question is answerable. This sounds like an appropriately quietist response. However, it faces two serious problems. The first is the dialectical issue that I used to motivate the task

35 Compare the use of rigidification in Dreier 2002 to make a similar point about possible metanormative theories whose content outruns their normative implications. 36 Against the attempt to resurrect quietism in a Kantian key, it is relevant that similar problems have been raised about Korsgaard’s theory: for example, Hussain and Shah “Metaethics and Its Discontents: A Case Study in Korsgaard.” (Ms. version 2.2.) have argued that Korsgaard’s theory is compatible with either realist or expressivist metaethics, thus suggesting that her view is compatible with either a metaphysical or psychological account of normativity. Korsgaard’s best hope of avoiding this problem is to defend the view mentioned above that the debate between metanormative realists and anti-realists rests on a mistaken understanding of the issues (for which compare her 2003). Besides being deeply problematic (see Hussain and Shah Ms. and 2006, and McPherson 2008), this strategy is unavailable to the quietist realist, who is an antagonist in the realism-antirealism debate. 37 Thanks to David Enoch for pressing this objection on the quietist’s behalf. 38 In the enormously helpful terms of Schroeder 2005, the divine command theorist offers a constitutive explanation. McPherson Against quietist normative realism MS 15

of explaining robust normativity. I suggested that the assumption that there is something that makes a standard fail to be robustly normative is required in order to understand Scanlon’s objection to naturalist metaphysical views like the second- order desire account of reasons. I then argued that such objections are crucial to motivate the quietist realist’s program. The second problem is deeper. Our quietist cannot merely argue that we cannot know what makes a standard robustly normative, as the mysterian of §1 might suggest. This epistemological view is compatible with any number of metaphysically expansive views about normativity. He must rather argue that there is nothing that makes one standard robustly normative. However, this would entail that there is nothing that explains the difference between reasons and schmeasons. The lesson to draw from this discussion is that the sort of objection that Scanlon applies against the second-order desire theory tacitly involves an external presupposition concerning the non-arbitrariness of the metaphysical nature of robust norms. Other practices make analogous external presuppositions. For example, on the usual potted story, Sixteenth Century talk of witches presupposed that people identified as witches had supernatural properties, such as magical powers conferred by infernal beings. This is why, in light of our disbelief in such supernatural properties, we take error theory about such witch-discourse to be plausible. Serious normative standards appear to be similar to witch discourse, and dissimilar from the rules of chess, just in involving such interesting external presuppositions. Consider another example: it is no objection to a claim about the normativity of the rules of chess that the contours of these rules are a matter of brute historical accident, as in fact they are.39 By contrast, I assume that quietist realists like Scanlon would join me in rejecting as absurd and repugnant the parallel idea that it might only be a brute historical accident that there are decisive reasons against killing humans for sport. If my proposal that robust normativity is an external presupposition is correct, then quietist realism will inevitably collapse into another metanormative theory. Thus, if the external presupposition turns out to be constitutive of our use of reasons-talk, then the would-be quietist realist’s denial that such external presuppositions can be vindicated suggests an error theory. If the external presupposition turns out not to be conceptually constitutive, then his rejection of such external presuppositions suggests the deflationary naturalism briefly introduced in §1. If he relents and accepts the existence of a metaphysical presupposition, then his view is no longer distinct from conventional naturalist or non-naturalist realism.

39 For example, Yalom 2004, 18 suggests that when chess was introduced to Europe from the Arab world in the Middle Ages, the rules were modified so that it was illegal to promote a pawn to a queen while one still possessed another queen. Yalom suggests that this change was made to avoid the appearance that the game condoned polygamy. While this change has since been reversed, I take it that instances of this sort of pattern of explanation of the rules of chess would in no way undermine the validity of those rules. McPherson Against quietist normative realism MS 16

Conclusions

Quietist normative realism is potentially beguiling. Doubts about the place of reasons (and other normative notions) in our broader conception of the world can seem profound and inescapable. This is disturbing, because we constantly engage in normative thinking in our everyday lives. Quietist realism suggests that such doubts about reasons are simply misplaced, and that our attentions should instead be directed towards substantive thinking about the reasons we have. I have argued here that at best Scanlon’s strategy for quieting those doubts prompts us to more carefully identify the locus of those doubts. I have also argued that the problem for quietist realism does not lie with Scanlon’s particular arguments. Instead, I have argued that the contrast between those reasons and an arbitrarily different normative standard cannot be explained in terms purely internal to our norms. Instead, our normative practices presuppose a commitment to a difference between those standards that can only be addressed by a metanormative theory. Metaphysically quietist realism is thus no easy answer to the metanormative task identified at the beginning of this paper. We must seek another way to understand how our normative and metaphysical commitments fit together.40

Works Cited Blackburn, Simon. Essays in Quasi-Realism. Oxford: Oxford UP, 1993. Boyd, Richard. “How to be a Moral Realist.” Darwall, Gibbard, and Railton 105- 136. Copp, David. “Moral Naturalism and Three Grades of Normativity.” Normativity and Naturalism Ed. Peter Schaber. Frankfurt: Ontos-Verlag, 2004. 7-45. ---. “Realist-Expressivism: A Neglected Option for Moral Realism.” Social Philosophy and Policy 18, 2001: 1-43. Darwall, Stephen, Allan Gibbard, and Peter Railton, eds. Moral Discourse and Practice. New York: Oxford UP, 1997. Dreier, James. “Metaethics and the Problem of Creeping Minimalism.” Philosophical Perspectives 18 (Ethics), 2005: 23-44. ---. “Metaethics and Normative Commitment.” Philosophical Issues 12 (2002): Realism and Relativism. Ed. Ernest Sosa and Enrique Villanueva. 241-263. Dworkin, Ronald. “Objectivity and Truth: You'd Better Believe It.” Philosophy & Public Affairs 25(2), 1996: 87-139. ---. “Replies to Critics: Zangwill.” Brown Electronic Article Review Service. Eds. Jamie Dreier and David Estlund. Posted 4/9/1997. Available at: http://www.brown.edu/Departments/Philosophy/bears/homepage.html

40 I am indebted to Sari Kisilevsky, David Enoch, William Fitzpatrick, Mark Van Roojen, David Plunkett, Mark Schroeder, Mark Newman, Sean Walsh, Jason Ford, the students in my 2008 Ethical Theory class, and audiences at UMD and the Central APA for helpful feedback on the ideas in this paper. McPherson Against quietist normative realism MS 17

Enoch, David. “Agency, Shmagency: Why Normativity Won’t Come from What Is Constitutive of Action.” Philosophical Review 115.2, (2006): 169-198. ---. “An Outline of an Argument for Robust Metanormative Realism.” Oxford Studies in Metaethics Vol. 2. Ed. Russ Shafer-Landau. Oxford: Oxford UP, 2007. 21-50. Finlay, Stephen. “Four Faces of Moral Realism.” Philosophy Compass 2.6 (2007): 820– 849. Fitzpatrick, William. “Robust Ethical Realism, Non-Naturalism and Normativity.” Oxford Studies in Metaethics Vol. 3. Ed. Russ Shafer-Landau. Oxford: Oxford UP, 2008. Foot, Philippa. “Morality as a System of Hypothetical Imperatives.” Philosophical Review 81.3 (July 1972): 305-316. Frankena, William K. “Main Trends in Recent Philosophy: Moral Philosophy at Mid-Century.” Philosophical Review 69.1 (Jan. 1951): 44-55. Gibbard, Allan. Thinking how to Live. Cambridge, MA: Harvard UP, 2003. Humberstone, I.L. “Direction of Fit.” Mind, 101.401 (January 1992): 59-84. Hussain, Nadeem “The Return of Moral Fictionalism.” Philosophical Perspectives 18 (2004): 149-187. --- and Nishi Shah. “Misunderstanding Metaethics: Korsgaard’s Rejection of Realism.” Oxford Studies in Metaethics. Vol. 1. Ed. Russ Shafer-Landau. New York: Oxford UP, 2006. 265-294. Jackson, Frank and Philip Pettit. “Moral Functionalism and Moral Motivation.” Philosophical Quarterly 45.178 (1995): 20-40. Kagan, Shelly. Normative Ethics. Boulder, CO: Westview, 1998. Korsgaard, Christine. “Realism and Constructivism in Twentieth-Century Moral Philosophy”. Journal of Philosophical Research, APA Centennial Supp. Philosophy in America at the Turn of the Century (2003): 99-122. ---. The Sources of Normativity. Cambridge, UK: Cambridge UP, 1996. Kramer, Matthew. Moral Realism as a Moral Doctrine. Oxford: Wiley-Blackwell, 2009. McGinn, Colin. The Problem of Consciousness: Essays Towards a Resolution. Oxford: Blackwell, 1993. Mackie, J. L. Ethics: Inventing Right and Wrong. Harmondsworth: Penguin, 1977. McPherson, Tristram. For Unity in Moral Theorizing. Princeton University Dissertation, 2008. Cited as 2008a. ---. “Metaethics and the Autonomy of Morality.” Philosophers’ Imprint 08.06 (July 2008). 1-16. www.philosophersimprint.org/008006/ Cited as 2008b. Moore, G.E. Principia Ethica. Cambridge: Cambridge UP, 1903. Nagel, Thomas. The Last Word. New York: Oxford UP, 1997. Olson, Jonas. “Reasons and the New Non-naturalism.” Spheres of Reason. Ed. S. Robertson. Oxford: Oxford UP, 2009. 164-82. Railton, Peter. “Moral Realism.” Darwall, Gibbard, and Railton 137-163. Rosati, Connie. “Naturalism, Normativity, and the Open Question Argument.” Nous 29.1 (March 1995). 46-70. Scanlon, Thomas. “Metaphysics and Morals.” Proceedings and Addresses of the American Philosophical Association 77.2 (2003): 7–22. McPherson Against quietist normative realism MS 18

---. What We Owe to Each Other. Cambridge, MA: Belknap, 1998. Schroeder, Mark. “Cudworth and Normative Explanations.” Journal of Ethics and Social Philosophy 1.3 (October 2005). 1-27. ---. Slaves of the Passions. Oxford: Clarendon, 2007. Shafer-Landau, Russ. Moral Realism: A Defence. Oxford: Clarendon, 2003. Smith, Michael. The Moral Problem. Oxford: Blackwell, 1994. Stoljar, Daniel. Ignorance and Imagination. Oxford, Oxford UP, 2006. Sumner, L.W. “Normative Ethics and Metaethics.”Ethics 77.2 (January 1967): 95- 106. Svavarsdottir, Sigrun. “Objective Values: Does Metaethics Rest on a Mistake?” Objectivity in Law and Morals. Ed. Brian Leiter. Cambridge, UK: Cambridge UP, 2001. 144-193. Tiffany, Evan. “Deflationary Normative Pluralism.” Forthcoming in Canadian Journal of Philosophy Supp. Vol. Velleman, J. David. “The Guise of the Good.” Noûs 26.1 (March 1992): 3-26. Wright, Crispin. “Truth in Ethics.” Truth in Ethics. Ed. Brad Hooker. Oxford: Blackwell, 1996. 1-18. Yalom, Marilyn. Birth of the Chess Queen. New York: Harper Collins, 2004. Zangwill, Nick. “Comments on Dworkin.” Brown Electronic Article Review Service. Eds. Jamie Dreier and David Estlund. Posted 12/2/1996. Available at: http://www.brown.edu/Departments/Philosophy/ bears/homepage.html Zimmerman, David. “Meta-Ethics Naturalized.” Canadian Journal of Philosophy X.4 (December 1980): 637-662.